Music Emotion Visualization through Colour
Januka Dharmapriya, Lahiru Dayarathne, Tikiri Diasena, Shiromi Arunathilake, Nihal Kodikara, Primal Wijesekera
Abstract
The possibility of retrieving colours relevant to musical emotions is an emerging multidisciplinary research concept. It can be considered as the initial starting point for music visualization. But due to the novice nature of previous findings and limited availability of emotionally annotated musical databases, obtaining accurate colours for emotions is a computationally challenging task. The purpose of this research is to obtain the most suitable emotional colour for a given song segment considering Russell's Circumplex Emotional Model. MediaEval Database for Emotional Analysis of Music (DEAM data-set) has been annotated in previous studies considering the aforementioned emotional model. In this research, a linear regression approach was used with the WEKA machine learning tool for the DEAM data-set. Effectiveness of the results compared with several linear regression models available in WEKA. Then the predicted emotion can use with Itten's colour model to obtain relevant colour for the emotion. According to the above comparison, the random forest linear regression approach provided the most reliable results compared to other models (accuracy of 81% for arousal and 61% for valence). This study unveiled that Russell's Circumplex model and Itten's colour model can effectively sync emotions in music and colours in arts. Therefore, it can be used to obtain effective music visualizations and musically synced artistic patterns considering emotions.